Professional Documents
Culture Documents
Escala FOUR PDF
Escala FOUR PDF
The Glasgow Coma Scale (GCS) has been widely adopted. Failure to assess the verbal score in intubated patients and the
inability to test brainstem reflexes are shortcomings. We devised a new coma score, the FOUR (Full Outline of UnRe-
sponsiveness) score. It consists of four components (eye, motor, brainstem, and respiration), and each component has a
maximal score of 4. We prospectively studied the FOUR score in 120 intensive care unit patients and compared it with
the GCS score using neuroscience nurses, neurology residents, and neurointensivists. We found that the interrater reli-
ability was excellent with the FOUR score (w ⴝ 0.82) and good to excellent for physician rater pairs. The agreement
among raters was similar with the GCS (w ⴝ 0.82). Patients with the lowest GCS score could be further distinguished
using the FOUR score. We conclude that the agreement among raters was good to excellent. The FOUR score provides
greater neurological detail than the GCS, recognizes a locked-in syndrome, and is superior to the GCS due to the
availability of brainstem reflexes, breathing patterns, and the ability to recognize different stages of herniation. The
probability of in-hospital mortality was higher for the lowest total FOUR score when compared with the lowest total
GCS score.
Ann Neurol 2005;58:585–593
Assessment of coma is a core clinical skill for physi- Attempts have been made to modify the GCS; how-
cians. Scales have been constructed to improve com- ever, most of these scales were more complicated, and
munication among health care personnel and also to were seldom used outside the country of origin.6,8
standardize examination of the unconscious patient. Others have suggested simplification of the GCS score
The most commonly used scale is the Glasgow Coma after documenting poor interobserver reliability in
Scale (GCS).1 Although the originators of the GCS traumatic brain injury.9 These concerns and prior at-
have reported data on practical reliability of the scale,2 tempts to design new scales strongly suggest a new
other studies have showed some difficulties in applica- scale is needed that could provide further neurological
tion by untrained nursing staff.3 Trained personnel detail in coma that is easy to use and that could predict
tend to apply the GCS better, although interpretation outcome. We sought to validate a new coma scale, the
of intermediate scores on the GCS remains difficult for FOUR (Full Outline of UnResponsiveness) score, and
emergency physician.4 thus compared it with the GCS.
Other shortcomings of the GCS have been recog-
nized. First, because many comatose patients are intu-
bated, the verbal component cannot be tested. Some Patients and Methods
clinicians use the lowest possible score; others extrapo- Description of the New Coma Scale
late the verbal response based on other neurological The new coma scale was named the FOUR score (Fig 1).
findings. Second, abnormal brainstem reflexes, chang- The FOUR score has four testable components, in contrast
with the GCS (Table 1). The number of components and
ing breathing patterns, and the need for mechanical
the maximal grade in each of the categories is four (E4, M4,
ventilation could reflect severity of coma, but the GCS B4, R4). (It is easier to remember than the GCS with its
does not include those clinical indicators. Third, the varying number of scores [E4, M6, V5] and is reinforced by
GCS may not detect subtle changes in neurological ex- the acronym.) The FOUR score detects a locked-in syn-
amination. More recently, a study in traumatic head drome, as well as the presence of a vegetative state where the
injury found lack of correlation between outcome and eyes can spontaneously open but do not track the examiner’s
GCS score.5 finger. The motor response is obtained preferably at the up-
From the Divisions of 1Critical Care Neurology and 2Biostatistics, Address correspondence to Dr Wijdicks, Mayo Clinic College of
Mayo Clinic College of Medicine, Rochester, MN. Medicine, Division of Critical Care Neurology, Mayo Clinic, 200
Received Apr 29, 2005, and in revised form June 11, 2005. Ac- First Street SW, Rochester, MN, 55905. E-mail: wijde@mayo.edu
cepted for publication July 17, 2005.
Published online Sep 26, 2005, in Wiley InterScience
(www.interscience.wiley.com). DOI: 10.1002/ana.20611
per extremities. The motor category includes the presence of have been validated previously and are reliable to assess alert-
myoclonus status epilepticus (persistent, multisegmental, ar- ness.11 Three brainstem reflexes testing mesencephalon,
rhythmic, jerklike movements), a poor prognostic sign in co- pons, and medulla oblongata function are used in different
matose survivors after cardiac resuscitation.10 The motor combinations. The clinical sign of acute third nerve dysfunc-
component combines decorticate and withdrawal responses. tion (unilateral dilated pupil) is included. The cough reflex
(We think this difference is often difficult to appreciate.) mostly becomes absent when both cornea and pupillary re-
The hand position tests (thumbs-up, fist, and peace sign) flexes are absent. Breathing patterns are graded. Cheyne–
Š Fig 1. Instructions for the assessment of the individual categories of the FOUR (Full Outline of UnResponsiveness) score (see Table 1). (A)
For eye response (E), grade the best possible response after at least three trials in an attempt to elicit the best level of alertness. A score of E4
indicates at least three voluntary excursions. If eyelids are closed, the examiner should open them and examine tracking of a finger or object.
Tracking with the opening of one eyelid will suffice in cases of eyelid edema or facial trauma. If tracking is absent horizontally, examine
vertical tracking. Alternatively, two blinks on command should be documented. This will recognize a locked-in syndrome (patient is fully
aware). A score of E3 indicates the absence of voluntary tracking with open eyes. A score of E2 indicates eyelids opening to a loud voice. A
score of E1 indicates eyelids open to pain stimulus. A score of E0 indicates no eyelid opening to pain. (B) For motor response (M), grade
the best possible response of the arms. A score of M4 indicates that the patient demonstrated at least one of three hand positions (thumbs-
up, fist, or peace sign) with either hand. A score of M3 (localization) indicates that the patient touched the examiner’s hand after a pain-
ful stimulus compressing the temporomandibular joint or supraorbital nerve. A score of M2 indicates any flexion movement of the upper
limbs. A score of M1 indicates extensor response to pain. A score of M0 indicates no motor response to pain, or myoclonus status epilepticus.
(C) For brainstem reflexes (B), grade the best possible response. Examine pupillary and corneal reflexes. Preferably, corneal reflexes are tested
by instilling two to three drops sterile saline on the cornea from a distance of 4 to 6 inches (this minimizes corneal trauma from repeated
examinations). Sterile cotton swabs can also be used. The cough reflex to tracheal suctioning is tested only when both of these reflexes are
absent. A score of B4 indicates pupil and corneal reflexes are present. A score of B3 indicates one pupil wide and fixed. A score of B2 indi-
cates either pupil or cornea reflexes are absent. A score of B1 indicates both pupil and cornea reflexes are absent. A score of B0 indicates
pupil, cornea, and cough reflex (using tracheal suctioning) are absent. (D) For respiration (R), determine spontaneous breathing pattern in
a nonintubated patient and grade simply as regular (R4), or irregular (R2), Cheyne–Stokes (R3) breathing. In mechanically ventilated pa-
tients, assess the pressure waveform of spontaneous respiratory pattern or the patient triggering of the ventilator (R1). The ventilator monitor
displaying respiratory patterns can be used to identify the patient-generated breaths on the ventilator. No adjustments are made to the venti-
lator while the patient is graded, but grading is done preferably with PaCO2 within normal limits. A standard apnea (oxygen-diffusion)
test may be needed when patient breathes at ventilator rate (R0). Figure reproduced with permission by Mayo Foundation.
N/N 20 0.48 0.66 0.43 0.50 0.70 0.50 0.75 0.84 0.72
N/R 20 0.84 0.87 0.79 0.54 0.75 0.85 0.66 0.87 0.77
N/NI 20 0.77 0.74 0.83 0.83 0.81 0.62 0.77 0.87 0.80
R/R 21 0.79 0.92 1.00 1.00 0.93 0.93 0.86 1.00 0.92
R/NI 19 0.89 0.83 0.84 0.81 0.85 0.89 0.79 0.91 0.89
NI/NI 20 0.87 0.69 0.89 0.81 0.81 0.79 0.69 0.80 0.75
Overall (95% CI) 120 0.78 0.80 0.81 0.78 0.82 0.77 0.77 0.88 0.82
(0.70–0.87) (0.72–0.88) (0.70–0.91) (0.68–0.88) (0.77–0.88) (0.69–0.85) (0.68–0.85) (0.81–0.96) (0.76–0.87)
FOUR ⫽ Full Outline of Unresponsiveness; GCS ⫽ Glasgow Coma Scale; N ⫽ nurse; R ⫽ resident; NI ⫽ neurointensivist; CI ⫽ confidence
interval.
The diagnoses of the patients selected for the study Cronbach’s ␣ showed a high degree of internal con-
were ischemic or hemorrhagic stroke (29 patients; sistency for FOUR score (␣ ⫽ 0.86 for the first rater;
24%), traumatic head injury (25 patients; 21%), cra- ␣ ⫽ 0.87 for the second rater) and the GCS (␣ ⫽
niotomy for brain tumor (13 patients; 11%), aneurys- 0.88 for the first rater; ␣ ⫽ 0.84 for the second rater).
mal subarachnoid hemorrhage (12 patients; 10%), Spearman’s correlation coefficients between GCS and
postanoxic-ischemic encephalopathy (10 patients; 8%), FOUR scores were high ( ⫽ 0.92 for both first and
spinal surgery (including trauma; 8 patients; 7%), sei- second ratings).
zures and status epilepticus (7 patients; 6%), other en- The rater agreement by level of conscious group was
cephalopathies (4 patients; 3%), central nervous system comparable between the two scales. The total weighted
infection (4 patients; 3%), acute neuromuscular disease scores of the FOUR score for the alert, drowsy, stu-
(3 patients; 2%), and miscellaneous acute neurological porous, and comatose groups were w ⫽ 0.57 (95%
conditions (5 patients; 4%). In two patients, eye open- CI, 0.24 – 0.90), w ⫽ 0.74 (95% CI, 0.53– 0.94),
ing and brainstem reflexes could not be tested reliably w ⫽ 0.75 (95% CI, 0.61– 0.88), and w ⫽ 0.70
because of posttraumatic eye swelling; thus, the best (95% CI, 0.57– 0.83), respectively. The total weighted
score was entered. Fifty-seven patients (48%) were in- values of the GCS for the alert, drowsy, stuporous,
tubated and mechanically ventilated. and comatose groups were w ⫽ 0.59 (95% CI, 0.33–
0.86), w ⫽ 0.69 (95% CI, 0.54 – 0.85), w ⫽ 0.72
Interrater Reliability of the FOUR Score (95% CI, 0.57– 0.86), and w ⫽ 0.69 (95% CI, 0.54 –
The overall reliability was excellent for both the FOUR 0.84), respectively.
score (w ⫽ 0.82; 95% CI, 0.77– 0.88) and the GCS The rater agreement by diagnosis of traumatic head
(w ⫽ 0.82; 95% CI, 0.76 – 0.87). The rater agree- injury was comparable between the two scales. The to-
ment was good to excellent for physician rater pairs. tal weighted scores of the FOUR score for the trau-
The highest degree of agreement was among the neu- matic and nontraumatic head injury groups were w ⫽
rology residents, and agreement was lowest among the 0.73 (95% CI, 0.57– 0.88) and w ⫽ 0.84 (95% CI,
neuroscience nurses for both scales (Table 2). The fre- 0.79 – 0.90), respectively. The total weighted values
quency of scores for each scale is shown in Figure 2. of the GCS for traumatic and nontraumatic head in-
Five patients became brain dead, and one patient had a jury groups were w ⫽ 0.71 (95% CI, 0.55– 0.86) and
locked-in syndrome. Myoclonus status epilepticus (in w ⫽ 0.84 (95% CI, 0.79 – 0.90), respectively.
all patients after cardiac resuscitation) was noted in five All nine raters agreed or strongly agreed (Likert
comatose patients, who also had no motor response to grade 4 or 5) with the five statements that addressed
pain. For the FOUR score, 82 of 240 (34%) observa- the clinical usefulness of the FOUR score.
tions had a maximal score of 16. The brainstem com-
ponent had the highest maximal scores (188/240 ob- Outcome Prediction of the FOUR Score
servations; 78%). The distribution of the scores for the Twenty-five (21%) patients died, and 72 patients
eye and motor components of the FOUR score was (60%) had a poor outcome (modified Rankin Scale,
comparable with the distribution with the GCS. A 3– 6).Table 3 presents the relations between total score
GCS total score of 3 was recorded on 34 occasions. Of and patient outcome for each of the two scales. Con-
these, only nine were scored at the lowest FOUR score sidering the FOUR scale total score, for every 1-point
of 0. In the remaining 25 instances, the brainstem re- increase in total score, there is an estimated 20% re-
flexes and respiration components provided additional duction in the odds of in-hospital mortality (odds ratio
information that allowed the raters to distinguish [OR] ⫽ 0.80; 95% CI, 0.72– 0.88). A 1-point increase
among the patients’ total scores (FOUR score of 1– 8). in total score is also associated with lower odds of poor
outcome defined as a modified Rankin scale of 3 or score of 9 (sensitivity ⫽ 0.75; specificity ⫽ 0.76) and
more (OR ⫽ 0.84; 95% CI, 0.77– 0.92). Both rela- a GCS total score of 7 (sensitivity ⫽ 0.80; specific-
tions remain after adjusting for age, sex, alertness ity ⫽ 0.80). Prediction was less accurate for poor out-
group, and diagnosis (traumatic vs nontraumatic). come (Rankin, 3– 6), with area under the curve at 0.72
With the GCS scale total score, for every 1-point for both scales.
increase in total score, there is an estimated 26% re- To visualize the relation between outcome and total
duced odds of experiencing in-hospital mortality under score, we examined scatterplots with superimposed lo-
the unadjusted model (OR ⫽ 0.74; 95% CI, 0.65– cal regression smoothers. We used a model-based
0.85). This relation remains after adjusting for age, sex, smoothing with generalized additive models (Fig 3).
alertness group, and diagnosis (traumatic vs nontrau- Although these relations are comparable, overall there
matic). A 1-point increase in total score is also associ- are some subtle differences that do exist. For example,
ated with lower odds of poor outcome (OR ⫽ 0.83; the probability of in-hospital mortality is higher for the
95% CI, 0.76 – 0.92). This effect is attenuated slightly lowest total FOUR scores when compared with GCS.
after considering the adjusted model (OR ⫽ 0.89; This is due to our observation that the patients with
95% CI, 0.76 –1.03). GCS scores of 3 may be further separated using the
Receiver operating characteristic curves were esti- FOUR scale. For both scales there appears to be a
mated to compare prediction of in-hospital mortality range of values above which the risk for in-hospital
between the two scales. The area under the curve was mortality is close to 0 (GCS ⬎ 8; FOUR ⬎ 12). Risk
equivalent for the two scales at 0.81. The sum of sen- for a poor outcome (Rankin, 3– 6) in general declines
sitivity and specificity was maximized at a FOUR total more gradually and is comparable between the scales.
FOUR score total 0.80 (0.72–0.88) 0.80 (0.68–0.93) 0.79 (0.68–0.93) 0.84 (0.77–0.92) 0.86 (0.75–0.97) 0.86 (0.75–0.98)
FOUR categoriesd
Eye 1.03 (0.68–1.58) 1.37 (0.79–2.35) 1.42 (0.82–2.48) 0.78 (0.55–1.10) 0.89 (0.58–1.36) 0.93 (0.60–1.43)
Motor 0.74 (0.47–1.17) 0.74 (0.44–1.24) 0.71 (0.43–1.20) 0.88 (0.58–1.34) 0.85 (0.54–1.35) 0.83 (0.52–1.32)
Brainstem 0.64 (0.41–1.02) 0.55 (0.33–0.91) 0.54 (0.33–0.91) 0.80 (0.48–1.36) 0.71 (0.41–1.22) 0.71 (0.41–1.22)
Respiration 0.76 (0.49–1.18) 0.73 (0.46–1.15) 0.71 (0.45–1.13) 0.90 (0.64–1.27) 0.91 (0.63–1.32) 0.91 (0.63–1.32)
GCS score total 0.74 (0.65–0.85) 0.73 (0.58–0.90) 0.72 (0.57–0.90) 0.83 (0.76–0.92) 0.88 (0.76–1.03) 0.89 (0.76–1.03)
GCS categoriesd
Eye 1.13 (0.63–2.03) 1.26 (0.64–2.49) 1.34 (0.66–2.73) 0.67 (0.40–1.12) 0.71 (0.39–1.29) 0.74 (0.40–1.37)
Motor 0.64 (0.45–0.91) 0.61 (0.41–0.90) 0.57 (0.38–0.87) 0.97 (0.69–1.37) 0.97 (0.68–1.40) 0.93 (0.64–1.36)
Vocal 0.65 (0.37–1.14) 0.65 (0.35–1.20) 0.66 (0.36–1.24) 0.82 (0.60–1.13) 0.89 (0.64–1.25) 0.92 (0.65–1.30)
a
Unadjusted logistic regression model.
b
Logistic regression model adjusted for age, sex, and consciousness group.
c
Logistic regression model adjusted for age, sex, consciousness group, and diagnosis (trauma vs nontrauma).
d
Estimates are based on a multivariate model, which includes all individual categories.
OR ⫽ odds ratio; CI ⫽ confidence interval; FOUR ⫽ full outline of unresponsiveness; GCS ⫽ Glasgow Coma Scale.
Discussion The FOUR score, unlike the GCS, does not include
The FOUR score is simple to use, includes the mini- a verbal response, and thus is more valuable in ICU
mal necessities of neurological testing in impaired con- practices that typically have a large number of intu-
sciousness, and specifically recognizes certain uncon- bated patients. In this prospective study using the
scious states. Our interobserver study is the largest FOUR score, all patients, except the two patients with
validation study of a new coma scale with 120 pairwise periorbital swelling from trauma, could be assessed re-
ratings involving 3 neuroscience nurses, 3 neurology liably. In contrast, the GCS, which uses a verbal score
residents/fellows, and 3 neurointensivists. The patient as one of the three components, was less useful in 48%
population studied is indicative of 3 months of admis- of patients because they were intubated. Most likely,
sions to ICUs and includes a diversity of acute neuro- the verbal agreement ( ⫽ 0.88) in our study is arti-
logical conditions, not exclusively trauma. ficially high because it merely requires that the patient
Prior validation studies on GCS have included 47 have an endotracheal tube inserted, which obviates the
neurosurgical patients6 and 64 neurointensive care pa- need to further examine the verbal response. This
tients,15 but 2 other studies involved less patients.1,3 would be expected because the verbal component has
The design of these studies, which involved the appli- been recognized as the least reliable component of the
cation of a single stimulus with multiple observers rat- GCS.16 The reliability of testing brainstem reflexes has
ing at the same time, is not reflective of how the scale rarely been studied in a large population of patients
is used in practice and eliminates an important source but was excellent among our raters. In one prior study,
of variability. In addition, we attempted to ensure that pupillary responses and oculocephalic responses were
the two ratings of each patient occurred as closely in tested in 28 patients, and fair interobserver agreement
time as possible to minimize the possibility that the was found for only the oculocephalic responses ( ⫽
patient’s condition had changed. Prior studies have in- 0.49).17 Examination of some brainstem reflexes has
volved newly graduated nurses or student nurses,3 re- been incorporated in the modified GCS (Glasgow–
search psychologists,15 paramedics,4 and occupational Liege Coma Scale). These reflexes included rapid neck
therapy graduate students supervised by a medical di- movements to obtain oculovestibular reflexes and eye-
rector of the neurointensive care unit.16 Our raters ball pressure to obtain oculocardiac reflexes. The inter-
were chosen because in practice they would examine observer agreement among 6 neurosurgeons testing 30
these patients, communicate their findings to each patients had a w of 0.69.18,19 We did not incorporate
other, and make decisions. these reflexes that could further jeopardize patients
The interrater reliability of the FOUR score and the who had additional spinal trauma and hemodynamic
GCS were of equivalent magnitude. This is remarkable instability. Abnormal respiratory breathing and ventila-
because the raters had only minimal experience with tory drive may have localizing value in comatose pa-
the FOUR score. In our study, the observer agreement tients, but we acknowledge important variables such as
was highest among residents, followed by neurointen- acute pulmonary disease and ventilator settings. Our
sivists, then nurses. There was perfect agreement study shows patterns of breathing can be easily mas-
among the residents in rating respiration and brainstem tered by physicians and interpreted satisfactorily by
reflexes, which is an important finding when commu- neuroscience nurses.
nication with the attending consultant is sought. Most discrepant scores were found among the neu-
roscience nurses in some categories of both scales. In GCS. The FOUR score recognizes a locked-in syn-
the neuroscience nurses pairs, the lowest agreement was drome and a possible vegetative state. The FOUR score
with grading eye responses in both the FOUR score includes signs suggesting uncal herniation. Attention to
and GCS and with the interpretation of brainstem re- respiratory patterns in the FOUR score not only may
flexes. Ratings of the eye responses are influenced by indicate a need for respiratory support in stuporous or
factors such as intensity of pain and loudness of voice, comatose patients, but also provides information about
fluctuating alertness in between ratings, or time spent the presence of a respiratory drive. The FOUR score
to obtain the response. The differences in observer further characterizes the severity of the comatose state
agreement may nevertheless indicate that the nursing in patients with the lowest GCS score. Finally, the
staff with more training should use the scales. How- probability of in-hospital mortality was higher for the
ever, the interrater reliability among nurses was still fair lowest total FOUR scores when compared with the
to good, and in none of the subcategories was a poor GCS.
interrater reliability found. Teasdale and Jennett1 noted more than 30 years ago
There are significant advantages over the GCS score. that “some may have reservations about a system which
The FOUR score remains testable in neurologically appears to undervalue the niceties of a full neurological
critically ill patients who are intubated. Intubation is a examination.” Our study shows that greater neurolog-
common procedure in the field, emergency depart- ical refinement in the FOUR score is valid when con-
ment, and ICU that invalidates one of the three com- fronted with a patient who has an impaired conscious-
ponents of the GCS. The FOUR score tests essential ness. Using our new scale, the examiner has an aid to
brainstem reflexes and provides information about describe these essential clinical features.
stages of brainstem injury that is unavailable with the Our study permits implementation of the FOUR